lamename

This is the correct answer. I'd add _multiple_ experts, because even they can disagree depending on the topic, which is expected and healthy. And even then, realize that experts can be wrong, but they're much less likely to be wrong than <random skeptic>.

I agree completely. Here's hoping that we can evaluate this once the open source project is actually released.

I wish they had just released it with this announcement, but from their open positions perhaps they want those filled before it goes live.

Time will tell.

The specific news is that they are open sourcing their Thousand Brains Project. [0]

Technical document: https://www.numenta.com/wp-content/uploads/2024/06/Short_TBP...

[0]: https://x.com/Numenta/status/1798406708987142171?t=PWgsdVsZl...

As an aside, I like the site author's Atoms/Fragments/Articles hierarchy for content on their website -- both in intent and the names themselves. I've been trying to come up with similar thresholds for my own work. Useful idea!

Employees are quite often given no notice of firing or layoff, despite it being discussed by managers or executives ahead of time without the employee present.

Is a Board firing the CEO typically conducted differently?

It may be that the way you like to think about things and the way others like to are different.

I find that quickly grasping a new architecture is easiest with a graph-based diagram first. Then code for details. All with the goal of internalizing the information processing steps. Not memory allocation per se.

In my mind, how the network implementation allocates memory is a different question.

But I think both of our desires just reflect our jobs, our interests, and simply how our brains conceptualize things differently.

This was the first cold-brewing device I ever encountered....as a guest at someone else's house. Didn't know it had a name. Thanks!

Thank you for saying this. When cold brew first came out, it was promoted as a brewing process that resulted in smoother (I'm guessing lower acidity) tasting coffee. Heating it up seemed natural, and its use in iced coffee seemed simply opportunistic. (In my experience at least).

Then it quickly caught on as a novelty, with nitro et al, and when I tell people I drink cold brew warmed I get looks of confusion or turned up noses.

But brew temp and serving temp are orthogonal.

What are the best ways to keep up with GOL developments?

Links you posted and hobbyist forums, formal research papers, or something else?

I'm interested. Scanning wikipedia, I'm missing the link with the importance of negative testing. Could you explain?

Like this or something different?

https://github.com/gzerveas/mvts_transformer

There are common answers but it really depends on: What questions do you want the data to answer?

Interesting. Can you explain how this is superior and/or different from traditional DSP filters or other non-tokenization tricks in the signal processing field?

Isn't it a given that classification would "outperform" regression, assuming n_classes < n_possible_continuous_labels? Turning a regression problem into a classification problem bins the data, offers more examples per label, simplifying the problem, with a tradeoff in what granularity you can predict.

(It depends on what you mean by "outperform" since metrics for classification and regression aren't always comparable, but I think I'm following the meaning of your comment overall)

I'm glad this is getting attention. If you spend time reading this literature, especially in the mental health space, there's a lot of sloppy ML published.

Unbelievably low sample sizes, little to no mention of careful cross-validation or other measures taken to prevent data leakage, etc.

In one case, a supervised ML model built for diagnosis using binary classification, with no training examples in the negative class, touting great true positive rates in the abstract.

I suspect part of the problem is people coming from other disciplines (ab)using machine learning who haven't been taught how to use it properly. However, even when the authors include a biostatistician/mathematician in the research, that doesn't mean the paper will be of high quality.

Agreed. Based on the title, I took the context of the article as generic data science/analytics in industry, not cutting edge bio or physics.

Do people really report 5 decimal places?

In my experience reporting models in industry, even 1 or 2 decimal places are irrelevant to the overall decision to be made, at least if the number is around "87%" as stated in the example.

Of course it's context dependent. If you're evaluating incremental improvements perhaps that warrants more careful precision.

After all, by definition each decimal place is 10x less important.

Have you see this one? https://jalammar.github.io/illustrated-transformer/

Have you seen the neural net series by Statquest? https://youtu.be/HGwBXDKFk9I?si=GxYKy1s996e6Q8-G

It builds a simple CNN and ends with a simple example of how multiple ReLU activation functions can approximate arbitrary curves.

While I appreciate your point, I think anyone who has spent sufficient time troubleshooting complex systems has dealt with similar types of problems, and can grasp the _spirit_ of the story.

In fact, I'd argue the quaint style of the story does geeks a favor: if it's appealing to normies, maybe they'll appreciate us technical folks' perspective a little more.

It's a meta-analysis, which means they evaluated 92 papers in the literature.

Is there a graphical explanation of the perceptual phenomenon the author is describing? I think I understand the physical stimulus he describes, based on diagrams in papers, but I havent found anything explaining this part:

> So translating the pupil around doesn't change the angle of the light rays, which means the image doesn't move.

Completely agree. Another issue plaguing some animal based neuroscience research is shoddy sample size. Understandable, given that some of the experiments are so damn challenging, but in the long run this does no favors to reproducibility.

That being said, for both of these issues, grants should award more money to fund the expansion in research protocol. Expecting any change without additional resources doesn't help.

Morbid indeed. Sounds like classic separation of Wanting vs. Liking in addiction

https://en.m.wikipedia.org/wiki/Reward_system#Wanting_and_li...

You might be interested in work being done by the Earth Species Project https://www.earthspecies.org/what-we-do/projects

I mostly agree. I started listening from the beginning and loved the deep leanring/CS/physicist/neuroscientist/philosopher guests.

I think he began running out of those people or just wanted to expand. Some of the new format guests are great, but I skip many episodes now, whereas before I enjoyed 99% of them.

Still, at its best it remains one of the most informative and insightful podcasts I know of, largely due to the guest expertise.

To be fair, what they mean is the way marketers and advertisers use "AI". Which is often without regard for any specifics.

I agree. I think it's imperative, but will come piecemeal...one industry at a time, and always too late.

Not quite "in a nutshell" but honestly the page + exploring references will get you far https://en.wikipedia.org/wiki/Consciousness

Alexey Guzey first came across my radar when his blog post on the institutions of Life Science[0] was posted here on HN. I was a neuroscience postdoc at the time. I found it partly true, with some experiences different from my own.

Then or soon after I saw he was working on New Science [1], an attempt at an alternative funding system for science outside of academia (my perspective, not his words). Having left academia myself, I found that intriguing and have always kept an eye on that space.

As for Guzey himself, who I've spoken with briefly on 1 occasion, I think he tries really hard to genuinely understand lots of things analytically and empirically, which is admirable. But anyone who does this will inevitably have blind spots that experts (or those with more life experience) spot immediately.

My hot take after meeting him was that he's very curious. He seemed like he's not afraid to be wrong, perhaps to a fault, and would rather just say what he thinks and have you tell him where he's wrong, than tiptoe with uncertainty.

[0]: https://guzey.com/how-life-sciences-actually-work/

[1]: https://newscience.org/